What problem does it solve? Jumping straight into code from training data leads to missed reusable work, outdated platform assumptions, and uninformed architecture decisions. This Skill performs structured research and discovery before building something new, producing a research brief that makes the actual build far more productive. ## Core Features & Use Cases - Three Depth Levels: Choose focused (one decision, ~30 min), wide (landscape understanding, 1-2 hours), or deep (exhaustive, 3-6 hours) research scopes. - Multi-Source Discovery: Scans local projects for reusable code, analyzes competitors, mines GitHub issues and forum sentiment, evaluates libraries, and audits platform capabilities (e.g., Cloudflare Workers, D1, Vectorize) against current docs rather than stale training data. - Structured Research Brief: Synthesizes findings into a markdown brief saved to .jez/artifacts/ covering competitive landscape, user needs, technical decisions, reusable assets, risks, and suggested build phases. - Use Case: Before building a markdown note app on Cloudflare, run deep research to discover that Workers AI toMarkdown, D1 FTS5, and Vectorize enable document import, full-text search, and semantic search — capabilities you would have missed from training data alone. ## Quick Start Ask the agent to do deep research on the product or feature you want to build, specifying a depth level such as focused, wide, or deep.
